Because we didn't get similar results from the results of two calibration of an 11667A power splitter equivalent source mismatch I was trying to confirm that our labs method is correct by using a third method published in Oct 1997 by Juroshek which uses a SOLT calibration of the VNA and splitter to obtain source match of the splitter used as an external port of the VNA.
An Agilent competitor in Germany has an application note which describes access to the source match term in service functions for specific analyzers. Our cal lab E8361a opt 14 PNA seems to have the same receiver access and cal viewer allows user to see the residual terms so I wanted to try to duplicate this on the PNA.
I tried connecting the input (PS1) to PNA port 1; the left port PS2 (label up and PS1 away from me) to Port 2, and performed 10M-18GHz SOLT cal at the third port of the splitter(PS3) using preset parameters using 85054B cal kit. The cal viewer source match term displayed a 12 db loss or about 1.6 VSWR. I took this as the output S22 not the equivalent output source match, and not what I was after.
Then I removed the reference receiver jumper, terminated the source and took the PS2 port to the Port 1 reference receiver input. The results of one and two port calibrations with unk thru were very similar.
Guidance on what I'm doing wrong would be appreciated as this seems to be an established method. Our lab implementation is terminated PS1 with PS2 and PS3 measured reflection and transmission on Port 1 and 2 then doing math. Those are the results I'm trying to validate. The other results are from Roseville standards lab cal and I think another method is used there.
The app note from the "Agilent competitor in Germany" appears to have an error in the description of the method. In the version of the app note that I have, the second paragraph on page 5 says: "The DUT is connected between the two test ports of the vector network analyzer. The DUT port of interest remains "open". It forms the reference plane and is the "new" port of the enhanced network analyzer." This setup is incorrect.
The correct setup is described later in the app note on page 8: "The input of the power splitter is connected to port 1 of the vector network analyzer. Port 3 of the power splitter is connected to the "R1 CH IN" input via a cable." R1 CH IN is the reference input for channel 1. An attenuator may be inserted in the reference path to avoid overloading the reference input. See the diagram in the app note.
The reference input for channel 1 may be accessible if your network analyzer has coax jumpers on the front or rear panel. Unfortunately the N5230C PNA-L that I would like to use has no jumpers as far as I can tell, so I don't think I can use the N5230C for the Juroshek method.
Have you been able to use the Juroshek method? Did you need to remove a coax jumper on the network analyzer?
To measure the equivalent source match of a two-resistor splitter you must measure the reflection coefficient of the output port of interest, with the other output port terminated and the input port terminated. Your result should be a magnitude of 0.25. Next, you must perform a transmission measurement from the output port of interest to the other output port, maintaining the termination on the input port. The result should be -12dB which converts to 0.25. Subtract your second result from your first result to get your answer for the one output port. Repeat the process in the opposite direction for the other output port. Of course in the real world these won't both be exactly 0.25 and this simplification is omitting the phase component. For the best results you'll want the magnitude and phase of both measurements. In order to subtract one from the other you will have to convert both into real and imaginary values, perform the subtraction, and convert the result back into polar form. See Google for the conversion details. Hope that helped.
This discussion is about an alternative method described by John Juroshek (NIST), that connects the splitter in such a way that it may be considered part of the network analyzer test set. Splitter source match is measured during the network analyzer error correction process. This method depends upon having a network analyzer that allows access to the reference path and to the error correction (cal) data. See: http://www.rohde-schwarz.us/en/service_and_support/Downloads/Application_Notes/?type=20&downid=793
Note that there is an error in the method description in the app note, as mentioned in the discussion.